Research on Short Text Classification Based on RoBERTa-TextRCNN

Abstract

Aiming at the problem of short text classification, this paper proposes a short text classification method for media data based on RoBERTa and TextRCNN. The semantic text vector representation is obtained through RoBERTa and used as the input of TextRCNN for training. After experiments on the THUCNews dataset, RoBERTa-TextRCNN’s accuracy rate reached 94.64%, which is 4.53% higher than TextRCNN and 0.62% higher than Bert+RCNN. The effect is better than other classification models, which proves its effectiveness in short text classification.

Topics & Concepts